• Chapter 10

Validation

• 10.1    Introduction

Validation against analytical solutions, experiments, and real incidents is the preferred way of documenting the range of scenarios for which a model system generates reliable estimates for relevant physical phenomena. Systematic validation is also essential for quality assurance: continuously monitoring the effects of changes in source code, compilers, operating systems. As such, Gexcon has implemented an automated framework for validation of the FLACS-CFD software. Currently, there are approximately 60 validation series included in the automated framework, with approximately 2,700 FLACS-CFD simulations (including a range sensitivity simulations). The cases range from small scale basic verification cases to large scale dispersion and explosion cases.

In this section, a small subset of the validation cases is presented as Validation Summary Sheets to highlight the validation status of FLACS-CFD.

• 10.2    Structure of Validation Summary Sheets

The validation results are presented as succinct validation documents (Validation Summary Sheets) that clearly communicate the FLACS-CFD performance for particular test cases.

• 10.2.1    Experiment details

The first page of the Validation Summary Sheet is used primarily to present the details of the experiment, highlighting the relevance for FLACS-CFD validation.

• 10.2.2    FLACS-CFD simulation results

  • 10.2.2.1    Performance metric parameters

Five performance metrics are provided for each campaign. These are

• •    the geometric mean bias (MG),

MG = exp(lnXp - EXO)                        (10.1)

• •    the geometric variance (VG),

VG = exp[(lnXp - lnX_O)²]                           (10.2)

• •    the fraction of predictions within a factor of two of observations (FAC2),

0.5 < Xp < 2.0                                  (10.3)

Xo

• •    the normalized mean square error (NMSE).

NMSE = ^(Xp二 ^Xo)2                         (10.5)

XpXo                             <   /

where Xp are the model predictions, X。and the observations and the overbar (e.g. Xo) is the average over the dataset. The number of data points used to calculate each metric (N) is also shown. Bounds are provided for each performance metric to show how they vary across the grid sizes considered for each validation exercise.

A perfect model would have MG, VG, and FAC2 = 1.0; and FB and NMSE = 0.0. Of course, there is no such thing as a perfect model. The individual performance metrics are coloured green if they fall in what is generally considered the acceptable range, i.e.:

• •    The fraction of predictions within a factor of two of observations is at least 50% (i.e. FAC2 > 0.5).

• •    The mean bias is within a factor of 2 of the mean (i.e. -0.67 < FB < 0.67 or 0.5 < MG < 2).

• •    The random scatter is about a factor of two to three of the mean (i.e. NMSE < 1.5 or VG < 4).

Note the above ranges are indicative only. Multiple performance measures should be applied and considered in any model evaluation exercise, as each measure has advantages and disadvantages and there is not a single measure that is universally applicable to all conditions.

• 10.2.2.2    Performance plots

Two plots types are shown to aid evaluation of model performance; parabola plots and scatter plots. Figures 10.1 and 10.2 present some examples of the scatter plots and their corresponding parabola plots.

• 10.3    Validation cases

Two-pagers for 50 experimental campaigns are presented, including two new cases for pool fires. A list of the cases is given under each subsection.

• 10.3.1    Discussion

All validation cases have been run with grids within the grid guidelines. Refer to section ”Recommended grid configuration” for discussion about the grid guidelines and a grid sensitivity assessment. To represent the much more rapid progression of detonation and have a consistent approach across both deflagration, DDT and detonation cases, we have reduced the averaging time of pressure to 0.1ms for the experimental results in all explosion cases (from 1.5ms previously). For deflagrations this is more conservative and may result in slightly increased underprediction in some of the validation cases, as the typical timestep in FLACS-CFD simulations will be longer than this. As the applied averaging time is different, validation performance cannot be exactly compared with previous validation, although in most cases the difference will be modest.

(a) Scatter plot, Xp = X。

(b) Parabola plot, Xp = X。

(c) Scatter plot, Xp > X。

(d) Parabola plot, Xp > X。

• (e) Scatter plot, Xp < X。                             (f) Parabola plot, Xp < X。

Figure 10.1: Examples of idealised data variation (low scatter) presented in scatter and parabola plots.

• Figure 10.2:    Example of data variation (high scatter) presented in scatter and parabola plots.

  【eeqj uweDulin

  
  
  
  
  
  

  • •    ETHANE. METHANE

  • •    HYDROGEN. METHANE

  • •    ETHYLENE

  • •    METHANE, PROPANE

  • •    METHANE

  PROPANE

  • •    HYDROGEN

  * MTOROGEK. N2

  • •    BUTMIEr ETHANE. METHANE, PEhTMNE, PR0P4NE ETHANE. METHANE, PROPANE

  
  

  • •    BUUNL PROMNE

  • •    METHANE

  • •    COZ, PROPANE

  • •    HYDROGEN

  
  
  

  • ETHANE. METHANE, N2

  • BENZENE, N_BU7ANE. N.HEXANE, N_PEN1ANE. N_PROPYLBENZENE, 0.XYLENE. TOLUENE

  • PROPANE

  ETHANE. HYDROGEN, METHANE. N2

  
  
  
  
  
  
  
  

  Experiment [kW/m2]

  
  
  
  
  

• Figure 10.3:    Scatter plot of validation cases: right).

  Explosion (top-left), Dispersion (top-centre) and Fire (top-

  
  

Table 10.1: Gas explosion validation cases.

• 10.3.2 Gas Explosion

Figure 10.3 (top-left) summarizes the presented explosion validation cases in a single scatter plot, while table 10.1 gives a summary of the individual experiments.

Performance is generally good, though we do see a spread in the results. There is a tendency for FLACS-CFD to perform better for large-scale experiments. This is because experiments at smaller scales are more sensitive to the modelling of quasi-laminar flame propagation and to radiative and convective heat losses. Different mixtures are included in the validation and the performance across different gas species is generally good. However, there are some cases of poor representation of very fast flames; for example, propane and ethylene overpressures are under-predicted. The under-prediction is especially evident for the ethylene cases. FLACS-CFD does not capture the flame instability types that become important at very high flame-speeds (approaching DDT), so the deviations in the results for propane and ethylene are somewhat expected.

In the performance plots, detonation cases are distinguished using a different marker style, as shown in the plot legend. Those are the hydrogen-explosion tests for which detonation was observed in the experiment. In the validation, FLACS-CFD is run with default settings for detonation (DDT ”AUTO” and DDT TIME -1), meaning that detonation is triggered for hydrogen explosions only when DPDX exceeds 1.

In addition to the validation against experimental data, the result of which is presented in the following subsections, FLACS-CFD version 22.2 has been tested for hydrogen explosion using a set of realistic fullscale scenarios. Simulations indicate a higher frequency of detonation events compared to the detonation frequency obtained in the simulation of hydrogen explosion validation cases. For hydrogen-explosion, thanks to the new detonation model in FLACS-CFD version 22.2, on average the user should expect an increase of predicted peak pressure of about 20% compared to the peak pressure obtained from simulating the same scenarios with version 22.1.

• 10.3.2.1    3D Corner Small Scale

3D CORNER SMALL SCALE

HYDROGEN, N2

Small scale hydrogen explosion experiments were performed in 3D-corner tests (an obstacle array of 37x37x37cm that is blocked on 3 sides and open on 3 sides). The tests considered 3 different obstruction densities, 2 ignition locations, and a range of H2 concentrations.

Location: Gexcon AS (NORWAY)

Year: 2004

Number of cases in series: 32

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 23 mm

REFERENCES

^[1]Renoult, J. & Wilkins, B. (2004), Hydrogen explosion safety phase 2 – small scale explosion experiments., Gexcon AS. Report number

GexCon-3-F46201-2

• 10.3.2.2    Bakerrisk

BAKERRISK

HYDROGEN, ETHYLENE

A set of large-scale unconfined vapor cloud explosion tests were performed in a congested region. The test rig consisted of cubic sections with 1.8 m long tubes located vertically serving as obstacles. Ethylene-air mixtures with concentration varying from lean to rich and lean hydrogenair mixtures were tested. The mixtures were ignited at the centre of rig just above the ground level. Pressure measurements were made using arrays of pressure gauges placed up to 91 m away from the rig ^[1][2].

Location:

Year:

Number of cases in series: 4

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 120 mm

REFERENCES

^[1]Kelly Thomas, Quentin Baker, Don Ketchum, Martin Goodrich, Max Kolbe (2003), Deflagration to Detonation Transition in Unconfined Vapor Cloud Explosions

All fuels                   10        1.4

-0.063

• 10.3.2.3    BFETS Phase 2B

BFETS P2B

ETHANE, METHANE, PROPANE

Phase 2B of the Blast and Fire Engineering Project for Topside Structures (BFETS) considered large-scale experiments in a test vessel with an internal volume of 50 m3. The vessel included varying geometric congestion and confinement. Experiments were conducted to study the effect of variation in gas concentration, ignition location, and water spray.

Location: Spadeadam, United Kingdom

Year: 1994

Number of cases in series: 14

SELECTION OF TESTS

Available measurements: Flame arrival time, Pressure

SIMULATION SETUP

Grids used: 500 mm

REFERENCES

^[1]Evans, J.A., Johnson, D.M. & Lowesmith, B.J. (1999), Explosions in full offshore module geometries (Work carried out under controct to the HSE, MaTSU/8847/3522), BG Technology. Report number R2422

^[2]Foisselon, P., Hansen, O.R. & van Wingerden, K. (1998), Detailed analysis of FLACS performance in comparison to full-scale experiments,

Christian Michelsen Research

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

FUEL                        N MG VG FAC2 FB NMSE

1

1

1.3

0.98

0.24

FUEL

ETHANE, METHANE, PROPANE

1

N    MG   VG   FAC2 FB     NMSE

370     0.53    2.6      0.58       -0.54         1.1

REFERENCES

^[1]Evans, J.A., Johnson, D.M. & Lowesmith, B.J. (1999), Explosions in full offshore module geometries (Work carried out under controct to the HSE, MaTSU/8847/3522), BG Technology. Report number R2422

^[2]Al-Hassan, T. & Johnson, D.M. (1998), Gas explosions in large-scale offshore module geometries: Overpressures, mitigation and repeatability.

^[3]Foisselon, P., Hansen, O.R. & van Wingerden, K. (1998), Detailed analysis of FLACS performance in comparison to full-scale experiments,

Christian Michelsen Research

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

All fuels

【peq】 UOAB-nE-s

REFERENCES

^[1]Evans, J.A., Johnson, D.M. & Lowesmith, B.J. (1999), Explosions in full offshore module geometries (Work carried out under controct to the HSE, MaTSU/8847/3522), BG Technology. Report number R2422

^[2]Al-Hassan, T. & Johnson, D.M. (1998), Gas explosions in large-scale offshore module geometries: Overpressures, mitigation and repeatability.

^[3]Foisselon, P., Hansen, O.R. & van Wingerden, K. (1998), Detailed analysis of FLACS performance in comparison to full-scale experiments,

Christian Michelsen Research

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

• BUTANE, ETHANE, METHANE, PENTANE, PROPANE

• ETHANE, METHANE, PROPANE

10.3.2.6 BFETS Partial Fills

ETHANE, METHANE, PROPANE

Available measurements: Flame arrival time, Pressure

BFETS PARTIAL FILLS

Location: BG Spadeadam (UK)

Year: 1994

Number of cases in series: 6

REFERENCES

A series of six explosion experiments were conducted. The explosion experiments were conducted in the full-scale explosion rig at BG Spadeadam, which is designed to be representative of an offshore module ^[1][2]. The experiments studied explosions produced by ignition of a homogeneous mixture of natural gas and air with the volume of the natural gas and air mixture ranging between 10% and 100% of the volume of the explosion rig. The six tests were carried out using a lengthened module of 28m x12m x8m. Variations in the experiments included different gas cloud size without mitigation techniques.

^[1]Evans, J.A. & Johnson, D.M. (1999), Gas explosions in offshore modules following realistic releases (Phase 3B): Data report for large scale partial fill experiments, BG Technology. Report number GRTC R3245

^[2]Johnson, D.M & Cleaver, R.P. (2002), Gas explosions in offshore modules following realistic releases (Phase 3B): Final summary report.

Report number R 4853

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

FUEL                         N    MG   VG    FAC2    FB    NMSE

ETHANE, METHANE, PROPANE                  374     1.1     1      0.99     0.1      0.032

1

1

FUEL

ETHANE, METHANE, PROPANE

1

N   MG   VG   FAC2    FB     NMSE

127      1.4      1.5       0.65       0.19         0.51

REFERENCES

^[1]Tomlin, G. and Johnson, D.M. and Cronin, P. and Phylaktou, H.N. and Andrews, G.E (2015), The effect of vent size and congestion in large-scale vented natural gas/air explosions, Journal of Loss Prevention in the Process Industries

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL                     N MG VG FAC2      FB NMSE

BUTANE, ETHANE, METHANE, PENTANE, PROPANE 53 0.98 – 1.4 1.3 – 1.5 0.77 – 0.83 0.16 – 0.19

REFERENCES

^[1]Shale, A.G. & Walker, D.G. (1996), Extended Modelling and Experimental Reseach into Gas Explosions (EMERGE) - Final report for medium and large scale explosion experiments. Report number R1327

All fuels 18     0.74 – 0.76

-0.46 – -0.3

10.3.2.9 Fh-ICT Lane Experiment

FH-ICT

HYDROGEN

The test setup consisted of a driver section that was a rectangular container (3 x 1.5 x 1.5 m). There was a squarespaced opening in the front wall of the driver section with a blocking ratio 0.1 (tests IA1, IA2, IA3) and 0.3 (tests IA4 and IA5). The container was followed by a "lane" which consisted of 2 parallel walls (thickness 4 cm) at a distance of 3 m with a length of 12 m and a height of 3 m. The walls were fixed with steel braces to ensure that they do not move during the experiments. The whole volume was filled with H2- air mixture. The hydrogen mixture was enclosed within the lane using a polyethylene foil (thickness 0.2 mm). The mixture was ignited at the rear side of the container with five distributed pyrotechnic igniters. Two vertical tubes (diameter 14 cm) were installed at a distance of 5 cm from the wall each in the middle and at the end of the lane for all tests except test IA1.

Location: Fraunhofer Institute for Propellants and Explosives (ICT), Germany

Year: 1984

Number of cases in series: 4

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 100 mm

REFERENCES

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL              N MG      VG       FAC2        FB         NMSE

3.5            0.38             0.37                1.3

10.3.2.10 Fireseal Pipe Penetration Blast Wall

FIRESEAL PIPE PENETRATION BLAST WALL

PROPANE, METHANE

Large-scale tests explosion tests were performed to study the behaviour and explosion blast resistance properties of passive fire protection from FireSeal AB. Standard Gexcon 50m3 module with geometry modifications was used. The main aim was to test the fire seal pipes on blast wall which was mounted using adaptor frame on the Gexcon module. The series of tests included some reference tests to fix the internal geometry layout/configuration that could give the desired explosion loading on the blast wall and pipe setup.

Location: Sutra,Burgen

Year: 11/04/2021 - 25/04/2021

Number of cases in series: 10

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 167 mm

REFERENCES

^[1]Brian A. Wilkins (2021), Explosion Testing of FireSeal Pipe Penetrations Seals and Blast Wall

10.3.2.11 FM Global 64m3 empty

FM GLOBAL 64M3 EMPTY

PROPANE, HYDROGEN, METHANE

The experiments were conducted in an empty 64 m3 chamber with dimensions 4.6m x 4.6m x3m with square vents of 5.4 m2 and 2.7 m2. The fuels used were methane, propane and hydrogen ^[1][2].

Location:

Year:

Number of cases in series: 18

4.6 m

‘Center

Ignition

Back |

Ignition、^

' ' Front

Ignition

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]Chao, J. Bauwens, C.R. & Dorofeev, S.B. (2011), An analysis of peak overpressures in vented gaseous explosions, Proceedings of the Combustion Institute

^[2]Bauwens, C.R. (2010), Effect of ignition location, vent size, and obstacles on vented explosion overpressures in propane-air mixtures, Combustion Science and Technology

Pressure

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Geometric mean bias

Experiment [barg]

10.3.2.12 FM Global 64m3 Obstacles

FM GLOBAL 64M3 OBSTACLES

METHANE, PROPANE

Vented explosion experiments were conducted using stoichiometric methane-air and propane-air mixtures in FM Global's large-scale explosion test chamber. The test chamber had overall dimensions of 4.6 x 4.6 x 3.0 m and an overall volume of 63.7 m3. A square vent with two possible surface areas, either 5.4 m2 or 2.7 m2, was located on one of the chambers vertical walls. The tests were focused on the effect of fuel, enclosure size, ignition location, vent size, and obstacles on the pressure development of a propagating flame in a vented enclosure. The dependence of the maximum pressure generated on the experimental parameters was analysed.

Location: Rhode Island, USA

Year: 2010

Number of cases in series: 10

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]Chao, J. Bauwens, C.R. & Dorofeev, S.B. (2011), An analysis of peak overpressures in vented gaseous explosions, Proceedings of the Combustion Institute

^[2]Bauwens, C.R. (2010), Effect of ignition location, vent size, and obstacles on vented explosion overpressures in propane-air mixtures, Combustion Science and Technology

0.3

0.69

10.3.2.13 FMGlobal Hydrogen Initial Turbulence

FMGLOBAL HYDROGEN INIT TURB

The Experiment is conducted in 64 m3 chamber with dimensions 4.6*4.6*3 with square vent of 5.4m2. Fuel used is Hydrogen with different concentration, initial turbulence and ignition location.

Location:

Year:

Number of cases in series: 15

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]Bauwens, C. Regis and Dorofeev, S.B. (2014), Effect of initial turbulence on vented explosion overpressures from lean hydrogen–air deflagrations, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL               N MG      VG        FAC2       FB        NMSE

• HYDROGEN

【peq】 co-^ra-DE-^

IO^-1 10°        IO¹

Geometric mean bias

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL               N      MG      VG       FAC2       FB        NMSE

【peq】 co-^ra-DE-^

• HYDROGEN

0.5         1.0

Experiment [barg]

HYDROGEN

22

QUG^Cro> u-」⁴-¹山 lu 090

Geometric mean bias

10.3.2.15 Hydrogen Balloon

HYDROGEN BALLOON

HYDROGEN

The experiment consisted of a 20 meters diameter polyethylene hemispheric balloon placed on ground and filled with stoichiometric hydrogen-air. The combustion was initiated by ignition pills at the centre of the hemisphere basement. Pressure dynamics were recorded using 11 transducers, installed on the ground level in a radial direction at distances from 2 to 80 meters away from the centre of the hemisphere basement at radii- R = 2.0, 3.5, 5.0, 6.5, 8.0, 18.0, 25.0, 35.0, 60.0 and 80.0 meters. Flame propagation was evaluated along the radial path between 45deg and 135deg from the point of ignition ^[1][2].

Location: Germany Year: 1983

Number of cases in series: 1

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 600 mm

REFERENCES

^[1]HySafe (2005), SBEP-V2: Fh-ICT Balloon Test

^[2]Hansen, O.R. & Storvik, I. (2005), SBEP-V2: Fh-ICT Balloon Test. FLACS-HYDROGEN CALCULATIONS. Report number GexCon-05-F46207-

Tech.Note no2

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL             N MG      VG       FAC2          FB           NMSE

1             1               0.036               0.045

【PQq〕UOAP-nE一s

10.3.2.16 Hydrogen Refueling Station

HYDROGEN REFUELLING STATION

HYDROGEN

The experiment in a mock-up of a hydrogen refuelling station was conducted jointly by Shell Global Solutions (UK) and the Health and Safety Laboratory (UK) in order to study the potential hazards and consequences associated with a hydrogen-air mixture explosion. The a€reworst-casea€K scenario of a stoichiometric hydrogen-air mixture explosion was offered for this simulation exercise.

Location: Shell Global Solutions (UK)

Year: 2007

Number of cases in series: 1

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 110 mm

REFERENCES

^[1]Makarov, D., Verbecke, F, Molkov, V., Roe, O., Skotenne, M., Kotchourko, A., Lelyakin, A.,Yanez, J., Hansen, O., Middha, P., Ledin, S., Baraldi, D., Heitsch, M., Efimenko, A. & Gavrikov, A. (2009), An inter-comparison exercise on CFD model capabilities to predict a hydrogen explosion in a simulated vehicle refuelling environment, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL             N MG      VG       FAC2         FB          NMSE

HYDROGEN

QUG^Cro> u-」⁴-¹山 LU 090

【PQq】 UOAQ-nE-s

Geometric mean bias

4 3 o.O-

2 o.

• HYDROGEN

10.3.2.17 HySEA Homogeneous

HYSEA HOMOGENEOUS

HYDROGEN

66 vented hydrogen deflagration experiments were performed in 20-foot ISO containers. 34 test were conducted with homogeneous and quiescent mixtures. The project objective was to improve the hydrogen safety for energy applications through pre-normative research on vented deflagrations ^[1][2].

Location: Gexcon AS (NORWAY)

Year: 2016-2018

Number of cases in series: 36

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 100 mm, 153 mm

REFERENCES

^[1]Skjold, T. (2018), Vented hydrogen deflagrations in 20-foot ISO containers

^[2]Skjold, T., Hisken, H., Lakshmipathy, S., Atanga, G., van Wingerden, M., Olsen, K. L., Holme, M. N., Turøy, N. M., Mykleby, M., van Wingerden, K. (2019), Vented hydrogen deflagrations in containers Effect of congestion for homogeneous and inhomogeneous mixtures., International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 16, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL          N        MG           VG            FAC2         FB      NMSE

〔6」pq- uo匚B-nLU-s

• HYDROGEN

10.3.2.18 HySEA Unipi

HYSEA UNIPI

HYDROGEN

To investigate vented hydrogen explosions in installations such as gas cabinets, cylinder enclosures, dispensers, and backup power systems. The experimental setup allows researchers to study the effect of internal obstacles (bottles), construction materials (including structural response), and the effect of mitigating hydrogen deflagrations by means of various venting devices. Experiments on small scale enclosure additionally provide information on the opening inertia of the vent panel at low hydrogen concentration and on the measurement of the structural response ^[1][2].

Location: University of Pisa (Italy)

Year: 2017

Number of cases in series: 9

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 90 mm

REFERENCES

^[1]Fineschi, F., Landucci, G., Schiavetti, M., Pini, T. & Carcassi, M. (2016), Small-scale enclosure Dimensions and design. Report number HySEA-D-2-01-2016

^[2]Fineschi, F., Landucci, G., Schiavetti, M., Pini, T. & Carcassi, M. (2017), Small-scale enclosure. Experimental campaign of hydrogen deflagrations for homogeneous H2 concentrations. Report number HySEA-D-2-03-2017

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL             N MG      VG       FAC2          FB           NMSE

【PQq〕UOAP-nE一s

10.3.2.19 JIP MEASURE AR8DR8 SD2 SD9

JIP MEASURE AR8DR8 SD2 SD9

PROPANE

Gexcon AS developed a Joint Industry Project, Modelling Escalating Accident Scenarios and the Use of Risk-reducing technology for Explosion safety (MEASURE), that included the conduction of large scale experiments to provide validation data for explosion modelling (e.g., FLACS-CFD). Here, we consider the tests using a configuration with type 8 congestion (an array of 8x8x4 pipes of 168.3 mm in diameter and 655 mm in pitch, offering approximately 17% volume blockage) for eight different separation distances (SD) ^[1][2][3].

Location: United Kingdom

Year: 2013-2017

Number of cases in series: 2

SELECTION OF TESTS

Available measurements: Flame arrival time, Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]Skjold, T. and Hisken, H. and Atanga, G. and Narasimhamurthy, V. D. and Lakshmipathy, S. and Storvik, I. E. and Pesch, L and Braatz, A. (2017), Final report 'JIP MEASURE'. Report number Gexcon-17-F46241-C-1

^[2]D. Allason (2017), MEASURE Data Report. Report number 1Q14RQR-11 Rev. 2-0

^[3]D. Allason (2017), MEASURE Data Analysis Report. Report number 1QI4RQR-19 Rev. 1-0

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

FUEL            N      MG       VG       FAC2          FB           NMSE

0.68

REFERENCES

^[1]Shale, A.G. & Walker, D.G. (1992), Modelling and Experimental Reseach into Gas Explosions (MERGE) - CEC Final Report.

^[2]British Gas (BG) (1992), Results from experiments with obstacle type A. British Gas Research & Technology Report.

^[3]British Gas (BG) (1992), Results from experiments with obstacle type B. British Gas Research & Technology Report.

^[4]British Gas (BG) (1991), Results from experiments with obstacle type C. British Gas Research & Technology Report.

^[5]British Gas (BG) (1992), Results from experiments with obstacle type D. British Gas Research & Technology Report.

^[6]British Gas (BG) (1993), Results from experiments with obstacle type E. British Gas Research & Technology Report.

^[7]British Gas (BG) (1993), Results from experiments with obstacle type CS. British Gas Research & Technology Report.

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

N MG       VG FAC2          FB             NMSE

19       1 – 1.1        1 – 1.1     0.95 – 1       0.0028 – 0.028         0.0056 – 0.048

19      0.99 – 1.1       1 – 1.1      0.9 – 1          -0.1 – 0.043            0.014 – 0.057

18    0.86 – 0.93     1 – 1.2    0.89 – 1         -0.26 – -0.1            0.04 – 0.13

57      0.98 – 1        1 – 1.1     0.91 – 1        -0.12 – -0.013           0.021 – 0.081

Pressure

All fuels

54

0.59

• ETHYLENE

• METHANE, PROPANE

• PROPANE

• METHANE

10.3.2.21 Mogeleg Branches

PROPANE

Available measurements: Pressure

MOGELEG BRANCHES

Location:

Year:

Number of cases in series: 6

REFERENCES

The campiagn consists of series of gas explosion experiments conducted in a channel 1.5m*0.3m*0.3m to study the effect of vegetation in flame propagation. The effect of vegetation both in terms of tree species, number of branches and foliage presence is investigated. The propaneair mixture was filled in the channel with ignition source at centre of the closed end. Four pressure transucers were used with one fitted at clsed end while other three placed along the rear wall.^[1]

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL             N      MG      VG        FAC2         FB          NMSE

①oup-」p> OCSEO^O

Geometric mean bias

0.35             0.66               0.54

.5.4.32 「 0.0.0.0. "E^o-- Co^ra-DE-^

Experiment [barg]

10.3.2.22 MOGELEG Channel

A series of 56 tests were performed in a 0.13m3 test chamber at the Gexcon test laboratory at Fantoft, near Bergen, Norway. The vessel is rectangular in shape and is 1.44m long, 0.3m high and 0.3m wide. The test vessel contained baffle plate obstructions for turbulence generation. The vessel is closed on all sides and at one end. The fuel mixtures were usually ignited near the centre of the closed end of the chamber. All explosion venting occurred at the end of the vessel furthest from ignition. A limited number of tests were performed however, in which the ignition source was mounted at the upstream face of one of the baffle obstructions. The internal geometry of the vessel could be varied by the insertion of baffle plate obstructions of 5cm, which were fixed to the walls and floor of the vessel. Due to the high reactivity of the gas mixtures to be used, and to maintain integrity of the test chamber, 2 obstacle configurations (2 baffles and 4 baffles) were used in the current experiments in addition to tests performed in the empty chamber without obstructions. 4 gases were used in the tests (H2, 3H2+CO, H2+CO, 3H2+N2) at different concentrations in air ^[1].

Location: Gexcon AS, Bergen, Norway

Year: 2001-2002

Number of cases in series: 33

Available measurements: Pressure

SIMULATION SETUP

REFERENCES

All fuels

1.8

0.72

10.3.2.23 Molkov

MOLKOV

PROPANE

The campaign consists of thirteen gas explosion tests conducted in cylindrical vessel of diameter 2m and length 3.5m. The vessel had a rectangular hole 1.7m high and 0.8m wide covered with sturdy paper at one of the edges. Propaneair mixture was used as fuel which was ignited by a point source located at 0.4m from covered edge on vessel axis. Obstacles were arranged in arrays in diametral plane of vessel. The internal pressure was measured by two pressure transducers. ^[1]

Location:

Year:

Number of cases in series: 14

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]V. V. Mol'kov and V. V. Agafonov and S. V. Aleksandrov (1997), Deflagration in a vented vessel with internal obstacles, Combustion, Explosion and Shock Waves

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL             N      MG      VG        FAC2         FB          NMSE

0.86             0.56               0.44

10.3.2.24 Park: Vented rectangular chamber

PARK

Campaign consists of series of explosion test conducted inside rectangular chambers with 700mm x 700mm crosssection and a large top-venting area of 700mm x 210mm. The height of chamber was varied from 200mm to 1000mm. Three different multiple obstacles with square, circular and triangular cross-sections were used with diameters/size 70mm and 100mm respectively. Methane-air mixture in all tests was ignited near the bottom wall. Pressure values were recorded with transducer mounted on top wall of chamber, 20mm from exit.

Location:

Year:

Number of cases in series: 20

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 13 mm, 25 mm, 33 mm

REFERENCES

^[1]Park, Dal and Lee, Y.s and Green, Anthony (2008), Experiments on the effects of multiple obstacles in vented explosion chambers,

Journal of hazardous materials

lO¹^

10.3.2.25 Shell 30m3 Harrison

SHELL 30M3 HARRISON

ETHANE, METHANE, PROPANE

A series of vented explosion tests has been carried out using 30m3 explosion chamber.One end of the chamber was provided with square orifice, a series of area reducing plates with vent area 1/2,1/4 and 1/8 of area of end face were used.

Fuel used are propane and natural gas ignited within the chamber at different ignition location i.e centre, front and end of the chamber.^[1].

Location:

Year:

Number of cases in series: 16

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 140 mm

REFERENCES

[1]A. J. HARRISON and J. A. EYRE (1987), External Explosionsa€^ as a Result of Explosion Venting, Combustion Science and Technology

① UUP_」Q>uc」山 LUO 山 D

10.3.2.26 Shell Hydrogen Repeated Pipe Congestion

SHELL HYDROGEN REPEATED PIPE CONGESTION

HYDROGEN

Hydrogen-air mixtures were ignited in a 3 m x 3 m x 2 m congested area at the Health and Safety Laboratory ^[1]. Three different levels of congestion were used by varying the amount of repeated pipes in the rig. Both rich and lean hydrogen mixtures were used.

Location: UK, Health and Safety Laboratory

Year: 2006

Number of cases in series: 7

SELECTION OF TESTS

Available measurements: Flame arrival time, Pressure

SIMULATION SETUP

Grids used: 133 mm

REFERENCES

^[1]Shirvill, L.C. & Roberts, T.A & Royle, M. & Willoughby, D.B & Sathiah, P. (2019), Experimental study of hydrogen explosion in repeated pipe congestion Part 1: Effects of increase in congestion, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

FUEL              N MG      VG       FAC2        FB         NMSE

0.64            0.52

Pressure

2 2 11 【praq】 UOQ-nLU-s

10.3.2.27 Shell NaturalHy Repeated Pipe Congestion

SHELL NATURALHY REPEATED PIPE CONGESTION

HYDROGEN, METHANE

Hydrogen-air mixtures were ignited in a 3 m x 3 m x 2 m congested area at the Health and Safety Laboratory ^[1]. The congested area consisted of repeated pipes. The amount of hydrogen in the experiments was varied from 0 (100\% methane) to 100 (0\% methane)..

Location: UK, Health and Safety Laboratory

Year: 2006

Number of cases in series: 5

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 133 mm

REFERENCES

^[1]Shirvill, L.C., Roberts, T.A., Royle, M,. Willoughby, D.B, Sathiah, P. (2019), Experimental study of hydrogen explosion in repeated pipe congestion Part 2: Effects of increase in hydrogen concentration in hydrogen-methane-air mixture, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

• •    HYDROGEN, METHANE

• •    METHANE

• •    HYDROGEN

• •    Deflagration X Detonation

10.3.2.28 Solvex

METHANE, PROPANE

The campaign consisted of gas explosions in large scale enclosure with size 10m*8.75m*6.25m and vent located at centre of the front wall. The experiments were performed with propane-air and methane-air mixtures iside the enclosure ignited by a point ignition source located at centre of the rear wall. The tests were conducted with empty encloure configuration, one or two rows of 0.5m diameter pipes.^[1][2]

Location: Shell UK

Year:

Number of cases in series: 8

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

REFERENCES

^[1]Molkov, Vladimir and Makarov, Dmitriy and Puttock, Jonathan (2006), The nature and large eddy simulation of coherent deflagrations in a vented enclosure-atmosphere system, Journal of Loss Prevention in The Process Industries - J LOSS PREVENT PROC IND

^[2]Vianna, Savio and Cant, R. (2010), Modified porosity approach and laminar flamelet modelling for advanced simulation of accidental explosions, Journal of Loss Prevention in the Process Industries

10.3.2.29 SRI Confined Tube

SRI-CONFINED-TUBE

HYDROGEN

The facility consists of a square section steel tube with an inside dimension of 38.1 cm and a length of 990 cm. The ignition end of the tube was closed with a steel plate and the opposite end was opened prior to ignition by piercing a latex rubber diaphragm stretched tightly over the end. Various obstacle configurations were used to promote turbulence of the flame propagation. Figure 1 is a cross section of the tube showing the layout of the obstacles. The steel obstacles were 6.35 cm thick and were attached to a steel floor that was 1.27 cm thick on top of 1.91-cm-thick risers. Blockage ratios of 0.32, 0.47, and 0.65 were obtained using blocks that had heights of 11.43 cm, 16.51 cm, and 22.86 cm. The blocks were spaced at regular intervals of 38.1 cm, 76.2 cm, or 152.4 cm for tests using 25 blocks, 13 blocks, or 7 blocks respectively. The first block was always 38.1 cm from the initiation end.

Location: DoE, USA

Year: 2002

Number of cases in series: 8

SELECTION OF TESTS

Available measurements: Flame arrival time, Pressure

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame arrival time

FUEL               N       MG      VG       FAC2         FB          NMSE

HYDROGEN

146

1.1           0.92             -0.16                0.11

Pressure

10.3.2.30 TNO 39m3

TNO 39M3

METHANE

The experiments were conducted in a concrete enclosure of size 4mx3.7mx2.6m with an internal volume of 38.5m3. Methane-air mixture was filled in the enclosure through three locations. Ignition source varied across the campaign with locations at rear, centre and near vent. Three pressure transducers were mounted inside the enclosure. Different vent configurations were used with single or multiple layers of polyethylene sheet and single plaster sheet plate as vent cover. Few tests were conducted with a single large obstacle placed at centre of the enclosure.^[1]

Location:

Year:

Number of cases in series: 6

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 173 mm

REFERENCES

^[1]W.P.M. Mercx, C.J.M. van Wingerden, H.J. Pasman (1992), Venting of Gaseous Explosions

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL           N      MG            VG           FAC2       FB        NMSE

①OUP_」B> OC^OEO^O

〔praq〕uo-lp-nuj-s

Geometric mean bias

4.1e+02               0            -1.7

0.20

0.15

0.10

0.05

• METHANE

10.3.2.31 Traffic Tunnel Square

Hydrogen explosion in a traffic tunnel (with and without cars). The experimental set-up consisted of a 78.5m long tunnel with a diameter of 2.4 m and a cross-sectional area of

3.74 m2. The experimental facility was a one-fifth scale mock-up of a typical tunnel for road transport. The explosive hydrogen-air mixture was located in a 10 m long region, filling a volume equal to 37 m3. Several experiments were carried out, using different hydrogen concentration. The experiments with 30% hydrogen volumetric concentration in air have been considered. The geometry set-up was either an empty tunnel or a tunnel with 4 vehicle models on the floor

centreline inside the hydrogen-air mixture. The vehicle models measured 0.94 m in length, 0.362 m in width and 0.343 m in height, representing typical real-vehicles at onefifth scale. The distance between vehicles was 1.52 m. The blockage ratio due to the presence of the vehicles was 0.03. Pressure transducers were mounted on the side wall of the

tunnel along its entire length. The ignition position was located in the middle of the tunnel for all experiments.

Location:

Year: 2009

Number of cases in series: 2

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

REFERENCES

^[1]Middha, P. & Kotchourko, A. (2010), CFD calculations of gas leak dispersion and subsequent gas explosions: Validation against ignited impinging hydrogen jet experiments, Journal of Hazardous Materials

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Pressure

FUEL              N MG      VG       FAC2         FB          NMSE

1               1               -0.21               0.047

Case name                 Type            Variable                                                           Metrics

N       MG         VG         FAC2            FB             NMSE

Table 10.2: Dispersion validation cases.

• 10.3.3    Dispersion

Figure 10.3 (top-centre) summarises the presented dispersion validation cases in a single scatter plot, while table 10.2 gives a summary of the individual experiments. Performance is very good across the different fuel types and ambient conditions. In general, the outliers can be explained by challenges modelling the variation of wind speed and direction in large-scale dispersion experiments. Measurements in experiments are typically across an arc-wise array of sensors. Narrow plumes sometimes miss all of the sensors on an arc, the resulting low concentration at the corresponding arc-distance causes problems in calculating the geometric mean and variance.

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL           N MG          VG           FAC2        FB         NMSE

CHRC

CO2, PROPANE

CHRC carried out wind-tunnel experiments on the dispersion of carbon dioxide (CO2), both with and without obstacles ^[1]

• ^[2] . One of the aims of the experiments was to produce validation data for the FEM3A dispersion model ^[3].The wind tunnel was an ultra-low-speed boundary-layer wind tunnel capable of simulating the constant stress layer of the atmospheric boundary layer. Airflow from the driving fans passed through a circular-to-rectangular transition to a 7 ft. high, 20 ft. wide and 80 ft. long working area in which the floor was covered with smooth rubber matting on which roughness elements were mounted.The obstacles, dike was square with an inner dimension of 63 cm and a wall height of 3.7 cm and the model tank was 31 cm in diameter with a hemi-spherical dome top and an overall height of 28.3 cm

Location: University of Arkansas, USA,

Year: 1988

Number of cases in series: 3

SELECTION OF TESTS

Available measurements: Fuel volume fraction

SIMULATION SETUP

Grids used: 6 mm

REFERENCES

^[2]Stewart, Jim and Coldrick, Simon and Lea, Chris and Gant, Simon and Ivings, Mat (2016), Validation Database for Evaluating Vapor

Dispersion Models for Safety Analysis of LNG Facilities: Guide to the LNG Model Validation Database Version 12

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL                 N MG VG       FAC2       FB        NMSE

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

15

0.093

2.6e+19             0.73           -0.14             0.19

Geometric mean bias

Experiment [vol%]

• 10.3.3.4    HySEA Inhomogeneous Release

HYSEA INHOMOGENEOUS RELEASE

HYDROGEN

66 vented hydrogen deflagration experiments were performed in a 20-foot ISO containers. 22 tests were performed involving either an empty container with only the frame inserted, or a pipe rack installed in centre position inside the container. The experiments involved release of hydrogen inside the container either from a circular pipe or from a cubical box located at floor centre of the container and above floor and later ignited with ignition location on the upper back wall ^[1][2].

Location: Gexcon AS (Norway)

Year: 2016-2018

Number of cases in series: 12

SELECTION OF TESTS

Available measurements: Fuel volume fraction

SIMULATION SETUP

Grids used: 100 mm, 150 mm

REFERENCES

^[1]Lucas, M, Skjold, T. & Hisken, H. (2020), Computational fluid dynamics simulations of hydrogen releases and vented deflagrations in large enclosures., Journal of Loss Prevention in the Process Industries

^[2]Skjold, T., Hisken, H., Lakshmipathy, S., Atanga, G., van Wingerden, M., Olsen, K. L., Holme, M. N., Turøy, N. M., Mykleby, M., van Wingerden, K. (2019), Vented hydrogen deflagrations in containers Effect of congestion for homogeneous and inhomogeneous mixtures., International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL        N MG       VG         FAC2             FB             NMSE

0.075 – 0.38

• 10.3.3.5    LLNL Falcon Tests

  LLNL FALCON

  
  

  METHANE

  
  

The Falcon tests ^[1] were performed to verify the effectiveness of vapour barriers in reducing the risk associated with the release and dispersion of LNG.The spill was distributed over the water pond by a 4-arms piping system. The spill pond was surrounded by a high fiberglass fence and upwind of the spill pond a @€xbillboard@€M structure was mounted to reproduce the typical turbulence generated by a storage tank.The three Falcon test cases are those which are regarded as benchmarks ^[2]. ^[3], ^[4][5]

Location: Franchman Flat (Nevada),U.S

Year: 1987

Number of cases in series: 3

Available measurements: Fuel volume fraction

SIMULATION SETUP

REFERENCES

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL             N      MG       VG        FAC2        FB         NMSE

KITFOX

CO2

Dense gas carbon dioxide (CO2) was released at ground level for longer time periods (continuous "plumes") and for shorter time (short-duration transient "puffs"), including both neutral and stable conditions. The desert surface, a dry lake bed known as Frenchman Flat, was artificially roughened using combinations of flat billboard obstacles in order to simulate the roughness of an industrial site and its surroundings at about 1/10 scale. The roughness elements were constructed from a plywood billboards and were catergorised as Equivalent Roughness Pattern (ERP, 2.4m x 2.4m) and Uniform Roughness Array (URA, 0.2m high x 0.8m wide). The roughness elements were installed along the cross-wind (y direction) and along-wind (x direction). O2 was released from a 1.5m x 1.5m area source, with a nearly constant emission rate. 84 concentration monitors were installed on the four downwind arcs (25, 50, 100, and 225m).

Meteorological instruments were installed on five towers with heights 24m (EPA), 4.9m (Met1), 8m (Met2), 4.9m (Met3), and 8m (Met4). Puff releases experiments comprised of 34, finite duration (20 seconds) tests. 13 tests were performed with URA and ERP in place, while 21 test with ERP removed [1]

Location: Frenchman Flat, Nevada, U.S.A.

Year: 1995

Number of cases in series: 51

SELECTION OF TESTS

Available measurements: Fuel volume fraction

SIMULATION SETUP

Grids used: 120 mm

REFERENCES

^[1]Hanna, S.R & Chang, J.C (2001), Use of the Kit Fox field data to analyze dense gas dispersion modeling issues, Atmospheric Environment

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

Geometric mean bias

Experiment [vol%]

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL             N      MG       VG       FAC2         FB          NMSE

10.3.3.8 Thorney Island

THORNEY ISLAND

BUTANE, PROPANE

In the Thorney Island continuous release experiments, a mixture of 32% R-12 and 68% nitrogen was released through a duct emerging from the ground and capped by 2 m diameter plate. The resulting outflow had zero vertical momentum and low radial momentum flow, comparable to a gravity current velocity ^{[1][2][3][4][5]}. Among the three trials, test number 46 was excluded because of the limited data due to changes in the wind direction during the experiment.

Location: Thorney Island, U.K.

Year: June 15,1984

Number of cases in series: 2

SELECTION OF TESTS

Available measurements: Fuel volume fraction

SIMULATION SETUP

Grids used: 380 mm

REFERENCES

^[1]Ermak, D. L., Chapman, R., Goldwire, H. C. Jr., Gouveia, F. J. & Rodean, H. C. (1988), Heavy gas dispersion test summary report. s.l.:

Lawrence Livermore National Laboratory., Lawrence Livermore National Laboratory. Report number ADA213880

^[2]Ivings M.J., Gant S.E., Jagger S.F., Lea C.J., Stewart J.R. and Webber D.M. (2016), Evaluating vapor dispersion models for safety analysis of

LNG facilities., Health & Safety Laboratory

^[3]Mauri, L. (2016), Evaluation of FLACS Performance Against the Model Validation Database, appendix 1. Report number GexCon-14-

F46033-C-1

^[4]McQuaid, J. & Roebuck, B. (1985), Large-Scale Field Trials on Dense Vapour Dispersion

^[5]Stewart, J.R., Coldrick, S., Lea, C.J., Gant, S.E. & Ivings, M.J. (2016), Guide to the LNG model validation database version 12.

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL                  N MG VG FAC2        FB         NMSE

0.98       1.6         0.67

REFERENCES

^[1]Ekoto, I.W., Merilo, E.G., Houf, W.G., Evans, G.H. & Groethe, M.A. (2012), Experimental investigation of hydrogen release and ignition from fuel cell powered forklifts in enclosed spaces., International Journal of Hydrogen Energy

^[2]W.G. Houf and G.H. Evans and I.W. Ekoto and E.G. Merilo and M.A. Groethe (2013), Hydrogen fuel-cell forklift vehicle releases in enclosed spaces, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Fuel volume fraction

FUEL              N MG      VG       FAC2        FB          NMSE

0.83            0.14              0.057

Table 10.3: Fire validation cases.

10.3.4 Fire

Figure 10.3 (top-right) summarises the presented fire validation cases in a single scatter plot, while table 10.3 gives a summary of the individual experiments.

The FLACS-CFD fire solver generally performs well for jet fires including horizontal and vertical jet fires, with or without crosswind, open, or impinging. Generally, parameters like flame length, flame trajectory, flame temperature, radiative and total heat flux compare well against experiments. Fire simulations for large horizontal non-impinging jet fires are shown to be dominated by significantly more buoyancy forces on the end part of the flame, causing the flame to bend off and rise up earlier compared to the experimental flame. The FLACS-CFD fire solver performs well for both steady flow rate and blowdown scenarios for various species including hydrogen.

10.3.4.1 Sandia Cryogenic Hydrogen Jet Fires

SANDIA CRYOGENIC HYDROGEN JET FIRES

HYDROGEN

The experiments were conducted at the Turbulent Combustion Laboratory of SNL in USA. The main aim of the experiments was to investigate the ignition and flame characteristics of cryogenic underexpanded jet fires. The analysed scenarios were concerned with hydrogen releases with temperature in the range 37-295 K and pressure 2-6 bar abs. The release temperature and pressure were maintained constant during each test and monitored upstream the interchangeable orifice of diameter 0.75 mm or 1 mm or 1.25 mm. The hydrogen was released vertically upward in the laboratory equipped with an exhaust gas collection system. The incident thermal radiation was monitored at 5 sensors located along the jet flame and at 0.2 m from the jet axis. Five tests with orifice diameter 1.25mm out of the entire set of experiments performed by SNL have been selected for the validation ^[1][2][3].

Location: Sandia National Laboratories, USA

Year: 2015

Number of cases in series: 5

SELECTION OF TESTS

Available measurements: Flame length, Radiative Flux

SIMULATION SETUP

Grids used: 100 mm

REFERENCES

^[1]Pratikash P. Panda and Ethan S. Hecht (2017), Ignition and flame characteristics of cryogenic hydrogen releases, International Journal of Hydrogen Energy

^[2]D.M.C. Cirrone and D. Makarov and V. Molkov (2019), Thermal radiation from cryogenic hydrogen jet fires, International Journal of

Hydrogen Energy

^[3]Muthusamy, D. (2019), Description of detailed scenarios for validation of consequence modelling tools - Gaseous hydrogen transport and use. Report number GEXCON-SH2IFT-WP3-2019-04_D3-2

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame length

FUEL              N MG      VG       FAC2        FB         NMSE

1               1               0.11               0.013

REFERENCES

^[1]Ekoto, I.W, Houf, W.G., Ruggles, A.J., Creitz, L.W, Li, J.X. (2013), Large-scale hydrogen jet flame radiant fraction measurements and modeling

^[2]I.W. Ekoto and A.J. Ruggles and L.W. Creitz and J.X. Li (2014), Updated jet flame radiation modeling with buoyancy corrections, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame length

FUEL             N MG       VG       FAC2         FB          NMSE

①uup_」e> UC^OEOa^

6 .8

Geometric mean bias

o o o o 5 4 3 2 【Lu〕UOA-nuJ-s

HYDROGEN

26

IO"¹

10°

o o o

6 4 2 ne/mxi u04qnE_s

Experiment [kW/m2]

REFERENCES

^[1]D.B. Willoughby and M. Royle (2011), The interaction of hydrogen jet releases with walls and barriers, International Journal of Hydrogen Energy

FLACS-CFD 22.2

DATE: FEBRUAR 16, 2023 POSTPROCESS QUANTITIES

Radiative Flux

FUEL              N MG      VG       FAC2         FB          NMSE

o² 1

1

1

0°

1

200      400

Experiment [kW/m2]

• HYDROGEN

REFERENCES

^[1]J. Gore and G. Faeth and D. Evans and Db Pfenning (1986), Structure and Radiation Properties of Large-scale Natural Gas/Air Diffusion Flames, Fire and Materials

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Radiative Flux

FUEL              N      MG      VG        FAC2        FB         NMSE

^uc^cra> u--GJUJO① o

7

1.

Geometric mean bias

Experiment [kW/m2]

REFERENCES

^[1]Barbara Joan Lowesmith and Geoffrey Hankinson (2012), Large scale high pressure jet fires involving natural gas and natural gas/hydrogen mixtures, Process Safety and Environmental Protection

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Flame length

– 0.96

-0.16 – -0.041

ETHANE, METHANE, N2

ETHANE, HYDROGEN, METHANE, N2

190 mm

100 mm

260 mm

180 mm

Radiative Flux

ETHANE, METHANE, N2

ETHANE, HYDROGEN, METHANE, N2

mm mm mm mm

• 10.3.4.6    LNG Foamglas Pool Fire

LNG FOAMGLAS

METHANE

The release of flammable gases, such as Liquefied Natural Gas (LNG) or Liquefied Petroleum Gas (LPG), may result in the formation of a flammable vapour cloud that is often dense, depending on the ambient conditions (particularly wind speed and direction). If this cloud encounters a source of ignition within the surrounding environment, a fire may occur. The characteristics of the resulting fire are dependent upon the release conditions and the environment into which the vapours are released. One test was conducted and the results of the test featuring the use of the passive material (FOAMGLAS® PFS System (Gen 2)) were compared to the baseline date obtained from the reference test.

Location: Centro Jovellanos, Asturias, Spain

Year: October 2013

Number of cases in series: 1

SELECTION OF TESTS

Available measurements: Radiative Flux

SIMULATION SETUP

Grids used: 125 mm

REFERENCES

• ^[1]    (2014), Vapour & Fire Control Testing of FOAMGLAS PFS System (Gen 2) on LNG

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Radiative Flux

FUEL             N      MG       VG       FAC2         FB           NMSE

1               -0.28               0.094

Geometric mean bias

Experiment [kW/m2]

• 10.3.4.7    Montoir LNG Pool Fire

MONTOIR POOL FIRE

METHANE

British Petroleum, Elf Aquitaine, Gaz de France, Shell and Total-CFP have collaborated to perform three LNG fire experiments which have been conducted successfully at Montoir, France, under different wind conditions in a shallow 35 m diameter bund. Large tests were essential to provide suitable data for use in models which would reduce uncertainties in the prediction of very large fires. Three fire tests were performed under different wind conditions during 1987.

Location: Gaz de France, Montoir de Bretagne methane terminal, France

Year: 1987

Number of cases in series: 1

SELECTION OF TESTS

Available measurements: Radiative Flux

SIMULATION SETUP

Grids used: 2187 mm

REFERENCES

spanDELKA and J. Moorhouse and F. TuckerR (1990), The montoir 35m diameter LNG pool fire experiments.

FLACS-CFD 22.2

DATE: FEBRUAR 15, 2023 POSTPROCESS QUANTITIES

Radiative Flux

FUEL             N      MG       VG        FAC2        FB         NMSE

METHANE

2

1

4.6

0.33             -1.2

Geometric mean bias

Experiment [kW/m2]

• 10.3.4.8    Gasoline 0.3m Pool Fire

  GASOLINE 0P3M POOL FIRE

  
  

BENZENE, N_BUTANE, N_HEXANE, N_PENTANE, N_PROPYLBENZENE, O_XYLENE, TOLUENE

To study the various parameters like burning rate, flame emissivity to further understand the amount of heat flux contributed from flames and its variation with respect to different pool diameters. The experimental setup consists of mild steel circular pans of 2 mm thick and 15 cm height.

Gasoline with mass burning rate of 77 gm/m2.s is considered as fuel for this study. Heat flux is measured at distance equal to pool diameter away from the pool centre on a vertical axis.

The heat flux gauge is traversed along the vertical axis. Location: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY, India

Year: 2013

Number of cases in series: 1

SELECTION OF TESTS

Available measurements: Radiative Flux

SIMULATION SETUP

Grids used: 18 mm

REFERENCES

^[1]Siddapureddy, Sudheer (2013), Characterization of Open Pool Fires and Study of Heat Transfer in Bodies Engulfed in Pool Fires

IO^-1 10°        IO¹

Geometric mean bias

• BENZENE, N_BUTANE, N_HEXANE, N_PENTANE, N_PROPYLBENZENE, O_XYLENE, TOLUENE

Table 10.4: Blast validation cases.

• 10.3.5    Blast

One, medium scale validation case that demonstrates the performance of the FLACS-CFD blast solver are included in the following subsection. Table 10.4 gives a summary of this experiment.

• 10.3.5.1    Urban Canyon

URBAN CANYON

The objective was to study the blast effects from medium sized high explosive (HE) and thermobaric explosive (TBX) charges detonated in an urban environment. The experiment was scaled 1/5. A simplified urban environment consisting of only four buildings was selected. The buildings were cubic and all dimensions optimized for the purpose of validation of numerical tools. The streets in the urban scenario were chosen to have the same width as the building dimensions, 2.3 meter. Four charge locations and two different heights of burst (0.2, 1.15 m) were used for the explosive charges ^[1].

Location: Sweden

Year: 2006

Number of cases in series: 8

SELECTION OF TESTS

Available measurements: Pressure

SIMULATION SETUP

Grids used: 200 mm

REFERENCES

^[1]Christensen, S.O. (2009), Urban canyon experiments and numerical calculations, The Norwegian Defence Estates Agency (NDEA). Report number Forsvarsbygg Futura Report 03/2009

①OUP-」Q> OCSEO^O

10      20

Experiment [barg]